In Situ Spatial Evolutionary Game Assay for Inferring Tumor Subclonal Payoff Matrices and Guiding Adaptive Therapy from Intact Biopsies
InSituGame solves a historic bottleneck in mathematical oncology: rather than guessing game parameters or using artificial 2D in vitro petri dishes, it estimates subclonal frequency-dependent payoff matrices directly from intact patient biopsy microarchitectures (10x Genomics Visium).
Removes extrinsic physical confounders (hypoxia CA9 and endothelial vascular proximity PECAM1) via multivariable spatial regression to isolate true cell-intrinsic game interactions.
Solves collinearity on the frequency simplex through an exact absorbed payoff formulation, ensuring numerical stability across thousands of spatial spots.
Formalizes Proposition 1, proving that static spatial snapshot transcriptomics rigorously maps the continuous velocity field of replicator differential equations.
Validates inferred subclonal trajectories against Monge-Kantorovich Wasserstein-2 mass transport across serial 3D histological tissue sections.
Demonstrates stable Hawk-Dove coexistence in Breast IDC (61.3% sensitive equilibrium) and Stag Hunt bistability in Cutaneous Melanoma (48.1% tipping point).
Simulates adaptive dosing schedules to suppress competitive release of drug-resistant clones, significantly extending simulated progression-free survival.